Doi: 10.1484/J.SEC.1.102502 SEM CLAS 4

1. INTRODUCTION

1.1. Reciprocal constructions have received much attention over the last decade, especially in the recent

-logical study by Nedjalkov in 2007.1 It is unfortunate that Nedjalkov’s enterprise did not dedicate even a single section to any Semitic language, ancient or

discussing aspects of these constructions in Akkadian, with comparisons to parallel constructions in other Semitic languages.2 This discussion will not exhaust

* Editions of Akkadian texts are quoted with abbreviations used in the Assyrian Dictionary, GELB, LANDSBERGER et alii 1956 [= CAD]. In general, when the examples in this paper appear in the CAD, I followed their translation, unless I either disagree with their proposals, or when I thought an alternative translation would be better for the purpose of the argument. For each example, if it is known to us, I will also indicate its time, and when neces-sary its location, as at times it will affect the discussion. The interlinear glosses are according to the Leipzig Glossing Rules. In adopting this system I had to use a different terminology than what is customarily used in the context of Akkadian, here are the various terminological adjustments: PST-preterite; SBJV-subordination marker; additional abbreviations: DSM-direct speech marker; ING-ingressive; INJ-injunctive particle; PRC- precative; ST-stative. I wish to thank Prof. Eran Cohen, Prof. Benjamin Foster, Prof. Eckart Frahm, Mary Frazer, Dr. Uri Gabbay, Prof. William Hallo, Prof. John Huehnergard, Nadine Pavie and Adam Strich for reading earlier versions of this paper and for their numerous suggestions, questions and corrections.

1. In addition, one should mention the following two volumes: FRAJZYNGIER, CURL 1999 and KÖNIG, GAST 2008, and the following articles: KÖNIG, KOKUTANI 2006 and EVANS et alii 2007.

2. At present there exist very few studies on Semitic languages which consider reciprocal constructions. For a study on Standard Arabic, see KREMERS 1997; on Biblical Hebrew, see JAY 2009; on Biblical and Mishnaic Hebrew, see BAR-ASHER SIEGAL forthcoming-a; on Modern Hebrew,

every aspect of this topic in Akkadian, but will elaborate on certain crucial aspects of reciprocal constructions, and will discuss several possible diachronic changes that we can trace in the history of Akkadian. As will

to recognize and then to understand better the various diachronic developments occurring in the history of Akkadian.

1.2. The corpus

Working on reciprocity in a extinct language raises

of reciprocal constructions per period or per dialect -

ization on such a restricted corpus is very limited. Therefore I took the Assyrian Dictionary [= CAD] as the primary corpus for this study, and added examples from various primary and secondary sources, including data from several grammar books. I conducted digital searches for English reciprocal expressions (such as “each other”) in the on-line volumes of the CAD. Once the different Akkadian reciprocal constructions were

and to examine the behavior of the expressions in Akkadian. While, in light of the corpus, it is obviously impossible to argue decisively for any generalization,

it is telling (statistically) if something is never found in this corpus. Similarly, if, for example, this study

and if it always behaves in the same way in this corpus, then the probability that this is true for all periods and all dialects of Akkadian is relatively high.

see SILONI 2001, 2008. In the context of Amharic, GOLDENBERG 1991, pp. 537-541 offers a survey of the pronominal constructions and AMBERBER 2002 discusses the various verbal constructions.

ELITZUR A. BAR-ASHER SIEGAL Notes on reciprocal constructions in Akkadian in light of typological and historical considerations*

24

In light of this, I decided to treat all periods and all dialects of Akkadian together, distinguishing between

-tions between periods or dialect. Working in such a procedure it is of course possible that I do not cover all the reciprocal constructions, and hopefully future studies that will focus on a more extensive corpus will

In addition I will bring examples from the periph-eral dialects alongside the “core dialects” as long as they behave similarly to all the dialects. I will note that an example is from a peripheral dialect only when the example shows a deviation from the standard with regards to issues related to reciprocity.3

2. THE TYPOLOGY OF VERBAL AND PRONOMINAL RECIPROCAL CONSTRUCTIONS

2.1. A proto-typical reciprocal construction is a gram-matical means to denote a mutual event(s). A clause which contains such a construction is said to substitute at least two propositions, thus we would say that (1a) is semantically equivalent to (1b):

(1) (a) Emily and Tony admire each other(b) Emily admires Tony and Tony admires Emily

When speaking of reciprocal constructions4 it is

and pronominal encoding of reciprocity. Reciprocal verbs can be morphologically encoded through different ways

morphological forms see NEDJALKOV 2007b, pp. 170-181). In Akkadian as is the case in the Semitic languages, it is marked by verbal templates, called in the Akkadian grammars stems.

Examples in (2)-(3) demonstrate the two types in Akkadian. The sentences in (2) illustrate the morpho-logical encoding, in which Akkadian has both verbs in the N-stems5 and T-stems. In (3) we include pronominal reciprocals of two types – “two-unit pronouns” (a-c) and “one-unit pronoun” (d):

3. Thus I will not comment on unique forms such as in (18d), but I will discuss the unique reciprocal pronouns from Susa (§ 3.5.1).

4. HASPELMATH which distinguishes between the semantic plane and the linguistic expressions, referring to the former as speaking about and the latter about reciprocal

Although I agree with the importance of this distinction, I will follow the common terminology in the literature and will speak about reciprocity for both planes.

5. However, as VON SODEN (GAG § 90e-g) noted, this func-tion of the N-Stem is very rare.

(2) a.I and you 1PL.PST.meet.RECPWe met, you and I (OIP 27 15:22, OA)

b. from day-GEN REL 1PL.PST.see.RECP-SBJV“Since the day we came to know each other” (Bagh. Mitt. 2 59 iv 12, OB)

c. [it]3MPL.PST.kiss.RECP=and 3MPL.PST.make friendship“They kissed each other and they formed a friendship”

d. COND father-NOM and son-NOM 3MPL.ST.angry.RECP“If a father and a son are angry with each other” (CT 39 46:75, NA)

e. ina GNin GN 3MPL.PST RECP“They fought in GN” (ABL 879:13 NA)

(3) a. NEG 3SG.DUR.see brother-NOM brother-ACC-3MSG.POSS“One person cannot see the other” (Gilg. XI 112, NA [SB])

b. afterwards brother-NOM to brother-GEN NEG 3SG.DUR.make.a.claim“Afterwards one will not make a claim against the other”(TCL 19 63:45, OA)

c. 6 73MPL-PST-kiss-ING brother- NOM brother-GEN “They began to kiss each other” (En. El. III132, NA [SB])

d. REL RECP 3MPL.DUR.lift=and 3MPL.DUR.swing“(The actors) who lift and swing each other” (CT 15 44:30, NA).

6. This form is in the N-stem. However, with this root reciprocity is expressed with a T-stem (see example 2c). As for the use of the N-stem in this context, it seems to be the ingressive use of this stem, and therefore the transla-tion: “They larger context where this line appears [“They entered

another in the Assembly” (En. El. III 130-132)]. Concern-ing the ingressive use of the N-stem, see inter alia VON SODEN (GAG 1952 [§ 90e-g]); KOUWENBERG 1997, p. 99. TESTEN 1998, p. 138 even argues that this is the original function of this stem.

7. The genitive here is not expected. It should be noted, however, that on another manuscript it is written logo-graphically (ŠEŠ-representation of the case.

25

While both strategies express reciprocity,8 they are substantially different from the point of view of the argument structure. In fact, as will become clear (§ 3.4), expressions of reciprocity with the so-called “reciprocal pronouns” are a subgroup within a larger group of strategies in which from the point of view of the sentence’s argument structure the sentences are “regular”, as the expressions conveying the reciprocals meaning (such as “one… another”) appear with the regular form of the verb, and occupy the regular positions of the subject, the object or other syntactic positions. In contrast, verbal reciprocals might be regarded as derivatives resulting from a process of detransitiviza-tion,9 characterized by two sets10 of participants which stand in a reciprocal relationship and occupy the same argument position while the other argument position remains “empty” (“Alex kissed Ruth and Ruth kissed Alex” vs. “Alex and Ruth kissed,” in which the latter does not have a direct object while the former does).

2.2. It is very common cross-linguistically that in the case of verbal reciprocals, in addition to “regular constructions” in which both participants are the subject, there is also the so-called “discontinuous construction.”11 In these constructions reciprocity holds between the subject set and the oblique set introduced

8. There are some studies (inter alia SILONI 2001, BAR-ASHER 2009a) which demonstrate some semantic differences between the verbal and the pronominal strategies. How-ever it is (almost) impossible to determine whether these distinctions hold in an extinct language such as Akkadian. Below in § 4 I examine whether some parts of the semantic typology are still relevant in the context of our study as well.

9. Similarly to NEDJALKOV 2007a and HASPELMATH 2007, I am using this terminology merely as a schematic description, without taking a position regarding an actual synchronic or diachronic derivation. In fact, BAR-ASHER 2009a demonstrates that we should not speak about an actual detransitivization as these are two separated items in the lexicon (in the sense of the linguistic “mental storage”). Instead it is advisable to speak about a structural relation-ship marked by the lexicon.

10. I am speaking about “sets” since the reciprocity can be a relation between groups and not just individuals. As example (20) below demonstrates, in the Akkadian sentence, “The people of Assyria and Babylonia mingle with each other” the relation is clearly between sets of individuals.

11. BEHRENS 2007 discusses the choice between the “regular” and the “discontinuous” constructions considering infor-mation structure parameters. An examination of this in Akkadian will require a more thorough study of the context of each example of these constructions in order to reveal the contrast between them.

by the associative preposition (itti in Akkadian) as in 4b, and not between the members of the subject as in (4a).12

(4) a. DN … at gate.of DN … 3MPL.PST.agree.RECP“They came to an agreement at the gate of DN” (CT 4 47a, OB)

b. COND shepherd-NOM GEN NEG 3SG.PST.reach.agreement“If the shepherd does not reach agreement

(CH § 57:50, OB).

It is worth mentioning that in the context of Akkadian, Streck (2003, pp. 82-86) assumes that the subject in reciprocals should comprise two agents. Consequently, he considers sentences in the discontinuous construc-tion and sentences with only a singular subject as later conceptually developments, which eventually resulted in the use of Gt verbs in non-reciprocal contexts. Approaching this material from a more typological perspective reveals that nothing is unique to Akkadian about discontinuous constructions, as even just the following examples from among the Semitic languages can demonstrate:

(5) 13 PN1 PN2when 3MPL.PST.be 3MPL.PTCP.quarrel.RECP PN1 and- PN2, 3mSG.PST.say to-him

1 ?(Jewish Babylonian Aramaic)PN1 with-me POSS.1SG 2SG.PTCP.quarrel.RECP“When PN1 and PN2 were quarrelling, PN1 said to him: ‘are you really quarrelling with me?’” (BM 85b)

(6) (Standard Arabic)

3MSG.PST RECP man-NOM.INDF with friend-GEN.INDF“A man fought with a friend”

(7) (Modern Hebrew)

DEF-boy-PLM 3PL.PST.kiss.RECP with DEF-girl-PLF“The boys and the girls kissed”

12. According to Siloni’s typology (SILONI 2001), such a construction is possible only in languages that, according to her theory, express reciprocity by the lexicon. In fact HASPELMATH 2007, p. 2093 proposed the following universal rule: “only verb-marked reciprocals allow discontinuous construction.” On this matter see also MASLOVA 2007, p. 337. The fact that in some languages discontinuous constructions are allowed with clitics seems to be problematic for this universal claim. BAR-ASHER 2009a, pp. 272-275 offered a solution to this problem.

13. t assimilates regularly

26

After presenting the two strategies of the encoding of reciprocity, I shall delve into each of the strategies separately, starting with the various pronominal construc-tions (§ 3) and then moving to some discussion about the extent of the verbal encoding (§ 4). In (§ 5) I will

and the verbal constructions, and examine a claim made in previous literature concerning their historical relation.

3. PRONOMINAL CONSTRUCTIONS

3.1. As is the case in other language families, there is no genetic relationship between the pronouns in the pronominal reciprocal constructions of each of the different languages. In general, these pronouns are transparent in various degrees in terms of their origins, as they are nouns at different stages of the process of grammaticalization, and used for describing mutual relations between sets.14 In fact, even within the same

being used simultaneously or, at least, frequent changes in the pronouns in the history of the language.15

As noted earlier, pronominal reciprocals regularly appear with the basic form of the verb (NEDJALKOV 2007b, p. 154),16these pronouns either take the position of one of the arguments or can appear in the place of other nouns in the sentence. Thus, with regard to basic sentences, pronominal reciprocal constructions can have the same argument structure.

Generally speaking, there are two types of recipro-cal pronouns: the one-unit pronoun co-referring with the plural subject (“to each-other,” “with each-other,” “each-others”); or the two-unit pronouns, preserving the iconicity of the reciprocal relations between the two sets participating in a reciprocal relationship (“one to another,” “one with the other,” “one the others”). As demonstrated in (3), Akkadian (like English for this matter) has both types of pronouns. The two-unit pronouns consist of a repetition of “brother”. The one-unit form is made up of in its various forms 17

14. For some preliminary collections of forms in the Semitic languages and proposals for their origin, see RUBIN 2005, pp. 22-23, and BAR-ASHER SIEGAL forthcoming-a.

15. This is, for example, the case in the History of Hebrew. See, .

16. Later in § 3.4.2 we shall discuss the conditions under which reciprocal pronouns may appear with the detransi-tivised verbal form.

17. Other forms of both types will be discussed further throughout the article.

3.2. The syntax of the sentences with the various types of pronouns

Akkadian shows an interesting distribution of the one-unit and two-unit pronouns.18 Based on a preliminary survey, while the one-unit pronoun may appear with its antecedent in the same sentence, the two-unit pronouns are used only when the antecedents are not part of the sentence. Thus, most often the members of the subject of the sentence with are the parties participating in the reciprocal relations. They appear either sepa-rately (8a) or as a group whose members participate in such relations (8b):

(8) (a) itti

COND raven-NOM and falcon-NOM with RECP ACC 3MPL.do.PST =and

(b) country-FPL-OBL to RECP say.DUR.3MPL“The countries say to each other” (Craig ARBT 1 26:8 NA)

Moreover, the one-unit pronoun never occurs as the subject of the sentence. It is either the direct object or it is the object of prepositions (for examples see § 3.6).

In contrast to this, in sentences with the two-unit pronouns, the participants in the reciprocal relations do

element of the two-unit pronouns is the subject and is in the nominative (slots of the other arguments in the sentence and appears in the appropriate case ( or ). Therefore, it is most often used in impersonal contexts (and very often in legal contexts), when the antecedents previously occur in the text (10), or when they are extraposed appearing in the absolute case (13-14). Thus, a sentence with an explicit subject [such as “the children saw one another”] was probably ungrammatical with two-unit pronouns. The following examples illustrate the use of the two-unit pronouns:

(9) a. brother-NOM toward brother-GEN something NEG 3SG.PST.have“None has a claim upon the other” (PBS 8/1 81:17, OB)

18. Previous work, such as GAG § 43b, mentions all pronouns together without any discussion about the distribution. The current survey is a preliminary study based only on examples from the relevant entries in the CAD. See also KOUWENBERG (1997, pp. 325-326) for some review of the strategies to express reciprocity in Akkadian.

27

b. brother-NOM against brother-GEN NEG 3MSG.DUR.claim“One will not raise a claim against the other” (BE 6/I 15:11, OB).

(10) you and enemy-2MSG.POSS 2MSG.DUR.get.into.

NOM

brother-ACC 3SG.DUR.destroy

An interesting example is found when these pro-

19 Such a rare example is found with the verb . In general this verb means “to put pressure on” and with a

) it has the meaning of “to exert oneself”. Thus in a reciprocal construction (31)

(11)expenses-ACC brother-ACC to brother-GEN NEG 2MSG.DUR.put.pressure“Do not exert pressure one on the other with regard to the expenses” (BIN 4 51:13-14, OA).

In all of the previous examples the verb is, as expected with , in the singular. In other sentences,

between the singular grammatical subject and the fact that reciprocity assumes multiple agents:

(12) a. brother-NOM brother-ACC NEG 3MPL.DUR.raise.a.claim“None should raise claims against the other”

b. brother-NOM brother-ACC concerning water.OBL NEG 3MPL.DUR.treat.opressively“One should not treat the other oppressively on account of the water” (TCL 7 23:29, OB).

This is a good example of what is known in the litera-ture (CORBETT 2006, pp. 155-160) as semantic agreement instead of a syntactic agreement (“the committee have

19. A different issue is when the reciprocity stands between the objects of causal relation. In such a case the common strategy for this function is with the preposition (“among”), for example: “you cause them to quarrel with each other” (RA 24,36).

met” vs. “the committee has met”).20 However, this also seems to be an example of a phenomenon not described yet in the typological literature on agreement (for a summary see CORBETT 2006, p. 160), as it is not that the reference of the subject is semantically plural (as is the case in “the committee have met”), but rather that the plurality is that of the events, and that in differ-ent events the same pronoun refers to a different entity among those who participate in the reciprocal relation.

Although the antecedents do not appear with the two-unit pronouns they can still appear as “nominative absolute,” and, as such, are no longer a part of the sentence:

(13) you and enemy-2MSG.PSG brother-NOM from brother-GEN

3SG.DUR.withdraw

(14)

since formerly I and you man-NOM to man-GEN ST.trust“From of old our relationship was such as one trusted the other” [lit. “From of old I and you trusted man to man”] (TCL 17 31:8f, OB).

The fact that “you and your enemy” (13) and “I and you” (14) are pre-posed and are not part of the main predication is indicated by the fact that the verb is in 3rd person singular and not 2nd person plural in (13) or 1st person plural in (14), as the verb agrees with the reciprocal pronoun . This can be clearly demonstrated by contrasting (14) to (15), in which the encoding of the reciprocity is verbal and not pronominal and, therefore, a plural subject:

(15) GN when I and you in GN 1PL.PST.see.RECP-SBJV“When you and I saw each other in GN” (PBS 7 108:10, OB).

While and seem to have a different distribution, in the sense that only the former is possible with an explicit antecedent, they, of course, do not exclude each other in the larger context. In (16), in order to form a poetic parallelism both strategies are used:

(16) brother-NOM brother-ACC NEG 3MPL.DUR.spare 3MPL.PRC.slay RECP“One should not spare the other, they should slay each other” (Cagni Era IV 135, NB [SB])

20. For a further discussion concerning this cross-linguistic phenomenon see BAR-ASHER SIEGAL forthcoming-a.

28

3.3. Two-units and one-unit pronouns from a diachronic point of view

In the previous section the syntactic functions of the various types of pronouns within the clauses was demonstrated as if they were both simultaneously available. Such an assumption is obviously valid since we encountered an example (36) where the two types co-occur. However, this assumption is not so simple since there is also a historical aspect to this distribution. It is quite evident that on the one hand the one-unit pronouns are used in a reciprocal meaning only from the Middle Babylonian / Assyrian periods, and on the other hand that the examples of two-unit pronouns are mostly from the Old periods and appear less frequently later on.21the later periods, however they appear mostly in classical texts written in SB (examples 3a, 3d and 16), or in legal texts as in the following marriage contract:

(17) PN PN2 [l-i] PN and PN2 wife-3MG.POSS GEN and in city-GEN

[ ] obligation-ACC of brother-NOM brother-GEN 3PL.DUR.do“PN and PN2, his wife, will serve each other’s obligations in both country and city” (TIM 4 45:8, MB).

With this information in mind we may move for a discussion about the origin of each of the pronominal strategy.

3.4. The origin of the two-unit pronouns

3.4.1. There is some evidence to support the claim that a repetition of any noun, besides , without a

tokens “implies another sentence with the reversed order of the same noun phrase” (NEDJALKOV 2007, p. 154).22 While it is known to happen with the word

“man” (18d), it occurs in all periods with many other words as can be demonstrated in the following examples:(18) (a)

king-NOM king-ACC in battle-GEN 3SG.DUR.defeat=and“One king will defeat the other in battle”

23

21. See BAR-ASHER SIEGAL forthcoming-a for a diachronic explanation for the shift from a two-unit to a one-unit construction.

22. For a similar phenomenon in other Semitic languages see JAY 2009, p. 7, n. 14 and BAR-ASHER SIEGAL forthcoming-a.

23. Whether this is example is relevant for our discussion see the end of the next discussion.

(b) 24surface-NOM over surface-NOM 3SG.DUR.higher“How much higher is one level (of water in the water clock) than the other level?”(TMB 26 nos. 50:3, 51:1, 52:4, OB).

(c) throne-NOM throne-ACC 3SG.DUR.overthrow“One throne will overthrow the other” (CT 27 25:24, NA)

(d) person-NOM to person-GEN thing NEG-3SG.PST.have“They do not owe each other a thing” (MDP 24 328: 8, LB, Achaemenid).

Sentence (7b), besides illustrating “reciprocity” without pronouns or special verbal form raises another issue that should be discussed further:

24. Sentence (18b) is peculiar in that both of the nouns which represent the participants of the reciprocal relation are in the nominative ( ), while the second participant is expected to be in the genitive, as it follows a preposition (This is the case both in 50:3 and in 52:4; in 51:1 it is a restoration). While this could be a

-enon in other sentences with reciprocal constructions, that both participants are in the nominative, while gram-matically one would expect that the second will be in the accusative, as the following examples illustrate:(a)

ambiguous(sign)-NOM ambiguous(sign)-NOM correspond.3SG.DUR“One ambiguous sign corresponds to another” (TCL 6 5 37f, LB, Seleucid)

(b) ...man-NOM man-NOM... NEG spare.3MPL.DUR“One man may not spare the other man” (Cagni Erra IV 135, NB [SB]).

The fact that this “grammatical error” recurs is striking. It is very likely that this is a result of the fact that two par-ticipants of the “reciprocal relation” are in many senses equally the “subjects”, as each of them at the semantic level occupies also the subject position, and in these example the semantics also affects the syntax. It should be noted, however, that (18b) is the strongest example among the three, as it is from a text usually dated, based on the

or a little later, and written in Old Babylonian. The other two examples are from later periods, in which the cases

– when the accusative –a ending is expected. Therefore only with more examples from the older periods this hypothesis can be strengthened. However, as for (a) note that this verb is in the Gt stem, thus the direct object is unexpected.

29

3.4.2. “Higher than” is of course not a reciprocal rela-tion, but a classic example of an asymmetric relation. However, cross-linguistically reciprocal pronouns are

relations between sets.” As, for example, is the case in the following English sentence:

(19) They are standing behind each other.

In (19), obviously, the one-unit pronoun does not express reciprocity but merely a relation between two sets, without specifying which of the sets occupies which position in this asymmetric relation. Thus, a natural conclusion would be that the reciprocity is only an interpretation resulting from the combination of

25 (WINTER 2007 reached some similar conclusions).26

A similar phenomenon is found with the Akkadian one-unit pronoun as the following example demonstrates:

(20) 2 two stars-GEN big-GEN after RECP 3MPL.DUR

the other” (Thompson Rep. 202 r. 4, NA27).

A similar example to (18b) is the following:

(21)

mountain-NOM NOM REL man-NOM after man-GEN 3MPL.PST.go

behind the other” (Wiseman Chron. P.74:11, LB).28

25. DALRYMPLE et alii 1998 survey the various logical relations that can be expressed by the so-called reciprocal pronouns in English. They examine what the various semantic

between the various sets that hold the relation expressed by the predicate.

26. According to this description we expect that this will be the case only with the pronominal encoding of reciprocity and not with the verbal. However already in OB we encounter the verb “to ride, to lie on top” in the Gt form and it is usually translated with the meaning of “to ride on top of the other” (besides the meaning of “to copulate”). The relation “on top of” is of course asymmet-ric, and therefore this is a similar phenomenon. However, this verb appears mostly in omens and the reality there is not completely clear whether it describes an asymmetric situation or that the two parties are mingled.

27. This document, however, is from a Babylonian rather than an Assyrian scholar.

28. In this case seems not to be a pronoun but an

describes actual people walking one after the other (see below § 3.5).

In light of this discussion, it is also not crucial to determine whether (18a) is a “real” reciprocal sentence, as most likely only one king defeats the other, but it is still relevant for our discussion since this is another

between sets. It should be noted, however, that in a negative sentence such as (18d) a reciprocal meaning is always attained (this is a necessary logical deduction):

(18d) person-NOM to person-GEN thing NEG-3SG.PST.have“They do not owe each other a thing” (MDP 24 328:8, LB, Achaemenid).

3.4.3. In the previous section we saw that in the history of Akkadian a repetition of any noun could express reciprocity. Thus naturally one could speculate the following stages in the grammaticalization of the pronouns:

(22) I. Originally, there was not a single word dedicated for expressing reciprocity. At this stage the word “brother” was used only in contexts where the meaning of “brother” is relevant.

II. Later, through a process of grammaticalization, semantically “bleached” and became a pronoun.

One can even imagine that this process began when was used in contexts similar to the following:

(23)house-NOM with house-GEN 3MSG.DUR.hostile brother-NOM brother-ACC

3MSG.DUR.kill“Family will turn hostile against family, brother will kill brother” (KAR 148:13, SB).

While this example is late, it shows a context where the original meaning of “brother” is relevant, but it could also be translated more generally as “one will kill the other”.

In the early stage, , was probably used only for people, and kept the gender distinction, as is the case in the following examples from OB:

(24) a. ;sister sister-ACC in secret-GEN word DSM

(F.PL) are saying secretly to each other” (Kraus AbB 1 135, 22)

b. sister-NOM against sister-GEN NEG 3MPL.DUR.sue“One woman will make no claim against the other” (CT 6 42b:9f)

following expected developments. It is used also in contexts with animals:

30

(25) [ ] COND eagle.PL.NOM brother-NOM brother-GEN 3MP.DUR.call=and“When eagles call each other” (CT 39, [Plate] 25, Sm1376:9, NA [SB]).29

Another somewhat expected development would

clear example with a feminine antecedent. It might be the case in example (26), from MB:

(26) PN PN2 [l-i] PN and PN2 wife-3MG.POSS GEN and in city-GEN

[ ] obligation-ACC of brother-NOM brother-GEN 3PL.DUR.do“PN and PN2, his wife, will serve each other’s obligations in both country and city” (TIM 4 45:8, MB).

Unfortunately the relevant part is restored, and the copyist did not draw the edges of the tablet on his copy, so there is no way of being certain whether there is enough space for two signs in order to have the feminine form instead.

Also relevant for this process of grammaticalization is what we encountered earlier in the lack of agreement in terms of number (12a-b).

Returning to the hypothesis in (22) concerning the process of the grammaticalization, the question is whether or not stage (II) annulled the possibilities available at stage (I). In other words, the question is: should we expect that, once there is a pronoun to express the reciprocity, there will be gradually fewer examples of repetitions of the same nouns?

, as noted above (§ 3.3) is rare in the later periods. In light of this, and based on our limited data, it is striking that in the later periods, there are not a few examples of the repetitions of the same nouns (see examples 18,21, sentences in n. 13). If our data indeed represent the development in Akkadian then we can conclude that the grammaticalization of the pronouns was independent from the use of the other nouns for this function. Moreover, the shift from two-unit to one-unit did not annul the grammatical basis of the two-unit pronouns, the repetitions of nominal expressions to express reciprocity. This could explain why even in the later periods the two-unit pronouns were still available, despite the grammaticalization of the one-unit pronoun.

29. Note that this is an example of two-units pronouns from a late period; but as noted this text seems to be written in SB.

3.5. The etymology of the various one-unit pronouns

3.5.1. Cross-linguistically, the elements of the two-units pronoun occasionally fused together (compare “one another” in English to “einander” in German). With Akkadian it seems unlikely that derived from a repetition of , but in other dialectal forms such a relationship can be more easily established. In Old Babylonian texts from Susa30of the following “one-unit pronoun”: or

(see CAD, A1 p. 193).31 The former is clearly a fusion of the “two-unit pronouns” into a “one unit pronoun.” In the case of the latter the elision of the second / / seems to be a result of haplology ( am > ).32 It is interesting to note how the same phenomenon happens in East Aramaic * ad ad > (with the addition of the plural marker). Maybe this is also the case in Greek: ἄλλη (another, f) ἀλλήλων (each other). In light of the observation that there is a diachronic relation between the two-unit construction and the one-unit construction (§ 3.3), the fusion of the two-units into one is the phonological representation of this development (see BAR-ASHER SIEGAL forthcoming [Hebrew] for the details of such a process).

3.5.2. As for 33 the variations with the /m/ in the Babylonian dialects, starting from the Middle Babylonian period, and the other forms in the Assyrian ones. While it seems likely that these forms are etymologically related to the component of the “two-unit pronouns” the origin of the various endings is obscure. There are two obscuri-ties with these forms: 1/ What is the origin of the /w/ and the /j/ in each of the dialects respectively? 2/ What is the nature of the -iš ending in this context, as it is usually an adverbial ending in Akkadian?

30. Susa’s dialect is considered as “peripheral” Akkadian, and it is assumed that its speakers had a different language as a substrate. However this is irrelevant for the typological discussion about the etymology of such pronouns.

31. Note that in the CAD the entry of these pronouns, as is the case with other nouns, are in the nominative. However, as “one-unit pronouns” they cannot appear in the subject position (“one loves the other” vs. “*each other loves”), as in fact the examples in the entry itself demonstrate. For more on these pronouns see VON SODEN 1933, p. 130, n. 1, MEYER 1962, p. 70, and SALONEN 1962, pp. 100-102. I wish to thank John Huehnergard for these references.

32. See also NEDJALKOV 2007, p. 201.33. I wish to thank John Huehnergard for discussing this

paragraph and for most of its associated bibliography.

31

GELB (1957, p. 104b) proposed that the origin of this pronoun is +išis a plural form of (compare with Old Assyrian

“sisters,” Syriac and “brothers” in Arabic).34 However, as

John Huehnergard has noted (p.c.), this is a problem-atic proposal since the -iš ending does not otherwise occur on plural forms.

However, it is possible to suggest that a historical /w/ in this word was not part of a plural marker, but rather a way to expand the root.35 In fact, in Hebrew we simi-

“friendship,”36 or the Arabic noun “friendship,” both with a similar expansion. In any case the development of

is expected in later dialects. Accord-ingly, the one-unit pronoun originated from a fusion of an expanded form of together with the -iš ending.

At this point, however, we should raise the issue of the -iš ending in this context. In the next section we will discuss whether is a pronoun or an adverb. At this point we should only note that if “friendship” is indeed the right etymology, then an analysis of this form as an adverb is more reasonable, assuming an original meaning of “brothers-like.” However, as will become clear it is evident that synchronically in Akkadian, in almost all periods, this is a pronoun, and, thus, we will have to assume a grammaticalization from an adverb to a pronoun – a process unknown in the literature about reciprocal pronouns (see NEDJALKOV 2007b, pp. 154-163).37

34. For a recent discussion about - as an external plural marker and a summary of the literature on the topic see HASSELBACH 2007, pp. 125-126. It should be noted that the phenomenon of the one-unit pronoun in a plural form is known from other languages. Earlier we saw the pronouns in the Late-Eastern Aramaic dialects, and one could mention the form ἀλλήλων as well.

35. VOIGT 2001, pp. 210-212 argues that the is part of the

whether this was part of the proto-Semitic root or a common way to expand bi-radical nouns.

36. This word appears only once in the Bible in Zechariah 11:14, and it is more common in Mishnaic Hebrew. In fact, we have an early rabbinic interpretation that demonstrates a relationship between “friendship” and “brother” (see 5:2).

37. It is of course possible that it developed from something similar to the forms from Susa ( ), discussed earlier, with the mimation of the singular form. However, since Susa is “peripheral” Akkadian, and has non-standard features, probably as it was written by non-native speakers, it is unlikely that its forms are the origin of the other dialects.

Due to this problem, it is worth considering an alternative etymology. This brings us back to the various spellings of the Assyrian alternative forms: and

.38 These forms suggest that originally the glide before the ending -iš was /y/ and not /w/.39 While one can suggest a phonological reasons for these forms in the Assyrian dialects,40 in light of the problem of the shift from an adverb to a pronoun, it is tempting to propose that there were three components to these pronouns: + , and that the middle component is the oblique ending of the dual (as is reconstructed for Proto-Semitic).

There are two motivations for this proposal:

I. As mentioned in NEDJALKOV 2007b, pp. 176-177, -

structions in other languages. This is of course not surprising since a reciprocal relation usually holds between two participants ( ). Among the Semitic languages, Arabic demonstrates a frequent use of the dual form in the VI form for reciprocal meaning, as the reciprocal relation is between two participants.41 Very rarely such a phenomenon is found also in early periods of Akkadian. In Old Akkadian there are examples of dual verbal agree-ment in the context of a reciprocal event (with the verb 42 and there are exam-ples of reciprocity with nominal dual endings from Old Babylonian when these morphemes were still somewhat productive:

(27) šarr-king-DU.NOM 3MPL.DUR.become.hostile.RECP“Two kings will become mutual enemies”

38. For the periodic/dialectal distribution see CAD, A1, p. 164a.39. Assuming that the vowel /e/ represents a contraction of

the diphthong / /.40. Intervocalic /w/ in the Middle Assyrian is usually writ-

ten , but there are examples such as for “man”. Thus theoretically, in Assyrian could represent . In fact Hecker, § 26a, e; 62a noted for some evidence of this phenomenon already in Old Assyrian. However, the fact that in these dialects is the normal, and forms such as are never found suggests an alternative etymology, according to which the /y/ is original, as the one proposed here.

41. Somewhat relevant to this is Sapir’s note that “the idea of reciprocity leads naturally to that of duality of terms involving mutual relationship” (SAPIR 1931, p. 110).

42. 20, p. 39-40.

32

II. The dual Semitic ending is for the oblique. As mentioned earlier (§ 3.2), with such a one-unit pronoun we expect that the forms will not be in the nominative as they do not take the subject position.

There are a few problems with this proposal that should be considered. In the history of Akkadian it is expected that the diphthong / / will contract to either / / or /the form , how can we explain the forms /

? It is possible that the ending - which follows the dual form, prevented this contraction as it starts with an /i/ vowel. It is, therefore, possible that such a contraction did not take place with the sequence / /. There is some evidence for such a prevention with the feminine gentilice ending / / which in some Akkadian dialects became - attesting to the lack of contractions in this environment (BUTTS 2010). Evidence for a similar glide are found in spellings such as

Another potential problem is the long / / in forms such as , which is represented in a spelling such as It should be noted, however, that such a spelling seems to be extremely rare.

The major problem with this proposal is of course the -iš ending. First, again we do not expect it to appear after a dual ending; second what is the nature of this ending in this context, as it is usually an adverbial ending?

However, one can propose a solution that will answer the various problems together, and also the co-existence of the forms with a / / and those with a / /, namely, that both solutions, should co-exist. As men-tioned above, following the proposal that the etymol-ogy is “friendship” is more reasonable that this form evolved in the context of the adverbial sense, and in the case of the discussed form, the meaning of this adverb was “together.” And, in fact, we demonstrated that similar abstract nouns, close in meaning, with the consonant / / are found in other Semitic languages. However, as mentioned earlier, the development adverb > reciprocal pronoun is less likely as it is unknown cross-linguistically. Therefore, I propose that this etymology (* ) should work for the adverb “together,” and that the dual ending (* ) for the pronoun “each-other.” If this is true, it is possible that these forms merged into one in the course of the history of Akkadian due to their semantic and phonological closeness. One could even speculate that, initially the pronominal forms did not contain the “adverbial” ending -iš (similarly to the forms from Susa), although I am well aware that

proposal. Thus I propose the following etymology for both

uses of these forms, before their mergence in meaning.

Accordingly the same phenomena happened in all dia-lects, but each of them eventually “picked” only one form.

* + Reciprocal pronoun:“each other”Adverb:“together”

The possibility that such to forms will become similar is not unlikely and in a footnote I propose that a similar phenomenon may have occurred in two Neo Eastern Aramaic dialects.43

Accordingly the original ending of the reciprocal pronoun 44

43. In The Jewish Neo-Aramaic dialect of Koy Sanjaq (de-scribed by MUTZAFI 2004) the regular reciprocal pronoun is In the context of the adverbial meaning of “together” it has two variants: either or Similarly, in The Jewish Neo-Aramaic dialect of Sule-maniyya (described by KHAN 2004), while the regular reciprocal pronoun is , in the context of the adverbial meaning of “together” again there are two variants , or most frequently Both andof the Eastern reciprocal pronoun or maybe even its older form (with both /d/ shifts into /l/ in Sulemaniyya and only the latter in Koy Sanjaq) Khan ( ., p. 259) explains the variations of the adverbial sense of “together” as a result of a reanalysis of the l at the beginning of as a preposition, hence its elision.

why it occurs with this preposition and not with the other ones, as appears with other prepositions and only in this function it is elided. Second, while it may explain the Sulemaniyya’s form it may not explain the form of Koy Sanjaq as d- is not a preposition. Moreover, the sociative meaning of the reciprocal meaning is usually attained with the sociative preposition. Thus in Sulemaniyya the expected preposition is min- ( ., pp. 346-347), as indeed it happens ( , p. 259, example [8]). Therefore I would like to consider that both and originally had nothing to do with the reciprocal pronoun , but derive from an independent adverb * (t) “in one” meaning “together”. Semantically this is very likely. Adverbs with the sense of together frequently derive from the cardinal number “one”, as for example is the Akkadian adverb išteniš (one + adverbial ending). Some support to this proposal may be found within an older Eastern Aramaic dialect, the Jewish Babylonian dialect where the sociative preposition is * > and it is very likely that it is another example of an adverb which became a preposition. If this proposal is correct then, and are the original ones and and

are the secondary, resulting from an assimilation to the reciprocal pronouns in their dialects respectively.

44. It is interesting to note that the ending of the one-unit pronoun in the Late Eastern Aramaic dialects: (Syriac), (Jewish Babylonian Aramaic), originated from a repetition of “one”, can also be explained as resulting from an additional agreement feature:

The ending can either be a vestige of a dual form ( ), or the regular Late Eastern Aramaic plural marker

33

The hypothesis that the origin of in the Akkadian form originated from an agreement marker is not an ad hoc, as in another Semitic language, Mehri, this phenomenon of an addition of an agreement feature is transparent, as the data represented in RUBIN 2010, p. 50-51 illustrate.

Occasionally reciprocity is expressed in Meheri, by a repetition of the word “one”:

(28) 3SG.throw.IMP stone one upon one“They throw stones on one another” (STROOMER 1999, 16, p. 2)

It is reasonable to assume that the more common one-unit pronoun is related to a repetition of

one-unit pronoun declines according to its antecedent, depending on whether the subject is plural or dual:

(29) 3MD.say.PRF to-RECP-DUdu“They (two) said to each other” (STROOMER 1999, 4, p. 17).

(30) then 3MP.understand.PEF RECP-3MPL“then they (pl) understood each other” (STROO-MER 1999, 59, p. 14).

once the one-unit pronoun grammaticalizes, it results with a dependency with the subject, expressed with an agreement.

3.6. : a pronoun or an adverb?

So far I have assumed that the forms are pronominal. These forms, however, have the

iš, known to be an Akkadian, and the same expression is undoubtedly used in Akkadian adverbially in the sense of “together, jointly”. Sentence (31) nicely illustrates the contrast between the uses of this form:

(31) together they=and RECP 3MPL.DUR.cause.trouble“They are together and (still) cause trouble for each other” (ABL 528 r. 5f. NA).

These facts are probably the reasons why others considered these forms to be adverbial.45

45. See, for example DELITZSCH 1889: 221, who discusses this form in the context of the adverbial ending -iš. The CAD characterizes it as an adverb, and so does BUCCELLATI 1996, p. 381.

As noted in the typological literature (NEDJAL-KOV 2007b, p. 162-163), in the context of reciprocal constructions the line between adverbs and pronouns is not so distinct, and, in fact, in some languages pronominal reciprocals can also function as adverbs (NEDJALKOV 2007, p. 163 brings Korean as an exam-ple). I believe, though, that the syntax should help us to distinguish between reciprocal pronouns (“each other”) and adverbs (“mutually”). Pronouns saturate the posi-tion of the “missing” argument (the “goal” in [32]), while adverbs do not and the slot for one of the argu-ments remains empty (the lack of “goal” in [33]). The

the identity of the goal.46 In English this distinction is also revealed by another syntactic behavior as the pronouns appear in the appropriate syntactic position (after the preposition), while the reciprocal adverb is not the object of any preposition and can appear in any slot available for adverbs:

(32) They gave each other gifts/ gifts to each other.(33) They mutually gave several proofs of their wit.

Since these are two different strategies to express reciprocity, it is not surprising that they do not exclude

-taneously in one sentence:

(34) They mutually gave each other the kiss of peace.

As for in the classical stages of Akka-dian (below we will discuss possible exceptions), it seems that this is an example in which the morphology is misleading (see above § 3.5). While it has an adverbial ending, it is quite certain that in the classical periods this “word” (in its various forms) functions as a indeclin-able pronoun (no overt cases) appearing as a one-unit

to (34) in which a pair of two-unit pronouns (such as the pair ) and appear together.

Moreover, functions like any other recipro-cal pronoun. Thus, with transitive verbs it occupies the position of the expected direct object:

(35) RECP 3MPL.DUR.pay“They will compensate each other” (Dar 321:29, LB).

While in the case of the direct object it is hard to demonstrate that is not an adverb, its appear-ance as the object of other prepositions reveals its pronominal nature:

46. See NEDJALKOV 2007b, pp. 161-163.

34

(36) to RECP NEG 3MPL.DUR.refuse“They will not refuse each other…” (TuM 2-3 2: 21)

(37)day-PL.OBL many.PL.OBL into heart-of RECP

weapon-PL.OBL-3MPL.POSS 3MPL.DUR.sharpen RECP 3MPL.DUR.cut.down“For many days they would sharpen their weapons at each other, they would cut each other down” ( 88, p. 126, NB).

Similarly, with the form from Susa (see above § 3.5.1):

(38) [l]child-PL.OBL-3MPL.POSS to RECPGEN NEG 3MPL.DUR.sue“Their children will not sue each other” (MDP 23 171:9-10, OB).

And in genitival constructions:

(39) people-PL.OBL of RECP 3MPL.DUR.kill“They are killing each other’s men” (ABL 645: 10f, NB).

(40) =mato help.of RECP 3MPL.PRF.trust=and“They trusted in one another’s help” (3R 7 i 43 Shalm. III, NA).

To complete the discussion of the pronoun the following example is crucial:

(41) people.of country.of Assyria country-of Babylonia with RECP

3MPL.DUR.mingle “The people of Assyria and Babylonia mingle with each other” (CT 34 39 ii 37, NA SB]).47

This example introduces the use of with predicates that are used by themselves to express recip-rocal events. The verb “to mix” in the N-stem has a symmetrical meaning (probably also with the addition of the ingressive aspect, therefore a better translation would be “they begin to mingle”).48 As is the case in other languages, with symmetric predicates it is possible to have, in addition to the use of the plural subject alone (“they disagree”) and to the discontinuous

47. GRAYSON 1972, p. 51 brings this line as an example of “some noteworthy phrases which are particularly common in this document”. However, he does not say why exactly it is noteworthy.

48. I wish to thank Benjamin Foster for emphasizing this to me.

construction (“she disagrees with him”), the combina-tion of both (“they disagree with each other”). From a theoretical point of view, this is similar to the dis-continuous construction discussed earlier (§ 2.2), as the object of the preposition itti is not occupying a grammatical slot dictated by the verb, and, therefore, it can be added despite the fact that it has the same reference as the subject.49

form that has a reciprocal meaning we do not encounter as the object without the associative preposi-

tion itti. From what we know from the cross-linguistic typology (see SILONI 2001 and BAR-ASHER 2009a) this is what we expect with detransitivized forms, as they do not have an object position. Had we encountered

without a preposition we would have charac-

in such a position strengthens our analysis that, despite its appearance with an adverbial ending, is not an adverb since it behaves perfectly as a pronoun. Accordingly, it is another example of a “one-unit” reciprocal pronoun that is mistakenly analyzed by modern scholars as an adverb, despite being a regular pronoun.50

3.7. A possible “one-unit pronoun” > “adverb” shift

As implied earlier, this analysis of pertains to the classical dialects of Akkadian. It is possible, however, that at some point in the long history of Akkadian was reanalyzed as a reciprocal adverb. In fact, evidence for such a process is extant in the late text of the Assurbanipal royal inscription from the 7th century BCE.51

49. For an elaboration on this theoretical aspect, see BAR-ASHER 2009a, pp. 266-270.

50. In the Semitic languages, another example is the pronoun in Jewish Babylonian Aramaic. For a discussion

about this and the support for this argument, see BAR-ASHER SIEGAL (forthcoming JBA § 8.1.2.3).

51. Another possible example is with the verb . According to CAD, in the Gt form it has the reciprocal meaning of “to accept each other” with the connotation of “making

, unbound to any preposition: PN PN2 “PN and PN2 made an agreement with each other” (UET 4 33:14, NB) and “After-wards they came into agreement” (TCL 12 14:9, NB [but dated to the reign of Sîn-šar-iškun, one of the last Neo-Assyrian kings]). However, it is very likely that these are perfect forms of the G-stem (see, for example,

“I have heard” [ABL 901:5, either NB or NA]) as a similar construction is found also with the G-stem: PN PN2 “PN and PN2 came to an agreement with each other” (VAS 6 331:7, NB). Thus it

Notes on reciprocal constructions in Akkadian in light of typological and historical considerations 35

is pronominal

preposition:

RECP 3MPL.DUR

itti.

PL.DUR RECP

people.of land.of Arabia one to one 3MPL.DUR

RECP

Interest

53

are from a late period.

53. translated to

“to release one

expressions.

RECP MPL.DUR RECP3MPL.DUR

4. THE SEMANTICS OF VERBAL RECIPROCALS IN AKKADIAN

BAR-ASHERSILONI DIMITRIADIS

events

HOPPER, THOMPSON

SILONI

PL.PST DEM.MSG ACC DEM.MSG PL

b.PL.PST RECP PL

SILONI

KOUWENBERG

GRONEBERG

55 If indeed

uC1taC2C2iC3

ACC REL PL.PRF RCP SBJV“T

MALONEY LOESOV

GEN

3MPL.POSS 3MPL.PST

55. KOUWENBERG

LIPI SKI

37

RECP “He and his groom stabbed each other with a sword in their belts” (Streck Asb 60 vii 36-37, NA [SB])58

The reciprocity here is clearly expressed by , but the question is whether it is also expressed by the verbal form. Reading as a Dt form will require either considering as an adverb, or understanding it as a case in which a Dt reciprocal form takes an accusative.59as seen earlier (§ 3.7), uniquely functions in this text as an adverb, and therefore it is possible to read the verbal form as Dt.

If we read it as a perfect D-stem (“have pierced”), then the is either the object of the sentence or is an adverb, and therefore lacks a preposition. We should remember that the pronoun behaves differently in the Assurbanipal royal inscription than in any other Akkadian dialect that we have encountered

substrate of another language, and, in any case, it is not the best text to gage the linguistic situation in Akkadian, with regards to reciprocity.

This is an example, I believe, of how sensitivity to typological considerations enriches our discussion

ancient text. Our examination exposed the components of the reciprocal construction. This understanding combined with the accurate analysis of as a pronoun elsewhere in Akkadian, and the knowledge that verbal reciprocals with T-forms typically have no direct objects,60 led us to our conclusion concerning

It should be emphasized that with regards to the last two examples I could not prove that the reciprocal readings are impossible. It is, of course, possible that Akkadian developed a reciprocal encoding unique

58. STRECK (1916, p. 61) translated it as a reciprocal “durch-bohrten sich,” but did not indicate whether it is due to

only or also due to the verbal form. For a new edition of the text see BORGER 1996, p. 59.

59. In both readings “their belt” should be taken as an adverbial accusative.

60. This is also relevant to the verb “to swear.” While semantically it can be used in the T-stem for mutual agreement with an oath (as is probably the case with the verb “to swear” for which we encounter a clear durative Dt form), the syntactic construction is with a direct object, literally meaning “evoking the name of the life of certain god.” Again all of the examples can be read as D perfect, and the reciprocity, if it existed, is received from the fact that both parties took an oath.

among the Semitic languages. However, our only intention was to add other considerations that should be included in this discussion.

Finally, it is worth mentioning in this context the different verbs for “to exchange.” This verb has a natu-ral reciprocal meaning, in which each of the “sources” (giver) is also a “goal” (the receiver). In the argument structure of verbs with this meaning there will be another argument, the object of the action that is expected to be the object. Therefore, we would not expect to encounter the T-stem with this verb.61 Indeed in Akkadian the regular verbs are and , both not in a T-stem. However, KOUWENBERG (1997, p. 326) proposed that there are examples in the Dt stem. According to what we saw in Akkadian and in the other languages, this is against our expectation.

analyzed as a D perfect form.62 that he mentioned, however, is doubtful.63

61. In Arabic the root has this meaning in different forms. In the case of the VI form, Arabic clearly shows a

which can be also passivized:(50)

3DU.PST.exchange Nasralla and-Olmert DEF-prisoner/PL “Olmert and the Hizbollah exchanged the prisoners”

(51) 3MSD.PASS.exchange DEF-prisoner/PL“The prisoners were exchanged”

This is clearly a different behavior than what we see in Hebrew and in Aramaic, as this is a use of a T-stem with a verb with high transitivity, and its inclusion of a clear direct object exhibits a different syntactic behavior. This evidence concerning Arabic invites us to look further for other verbs in Arabic with high transitivity, and to examine whether Arabic developed a true reciprocal marker, or

-cate in Arabic (probably developed by some analogy). This is another example of how we should be careful when we generalize from one language to another. It should be noted, however, that not all speakers of Standard Arabic

of different substrates. 62. In fact, such a construction is common with a perfect

form, as can be seen from the entry in the CAD. However, Von Soden AHw 1280 also analyzes some of the examples as Dt forms.

63. The problem with the example he mentions (p. 327) is that this verb seems ungrammatical without an object. Since this would be the only example of Dt it would be better to consider this as a corrupted text.

38

5. VERBAL VS. PRONOMINAL RECIPROCAL CONSTRUCTIONS

BUCCELLATI (1996, p. 381) considers verbal and reciprocal constructions in Akkadian to be similar, one synthetic and one analytic. Indeed, in most sentences, semantically, the two constructions are interchangeable but, as noted in the previous section, typological studies show distinctions between them, and occasionally they differ semantically as well. Buccellati’s approach led him to the claim that “the full reciprocal force of the Bt stems seems to have been lost in the later periods of Babylonian, for in these periods adverbs such as

‘together, mutually’ come to be used more and more either in place of, or next to, verbal forms of the Bt stem”.64

In fact, Buccellati is referring to two different phenomena: a/ a detransitivized verb with the one-unit pronoun , and b/ a transitive verb with a recip-rocal pronoun. He thinks that the frequent appearance of the two constructions indicates that the T-stems lost their reciprocal power in the later periods. It is very important to distinguish between the two phenomena. As long as the one-unit pronoun, , does not appear as an adverb with the detransitivized verb (and as we saw in § 3.6-7, it almost never does), it often follows the preposition itti “with” with the one-unit pronoun (see [41], and below [59]), or appears after another preposition. Thus, as noted earlier (§ 3.6), from a grammatical point of view, such an appearance of the detransitivized verb with the one-unit pronoun

is similar to the discontinuous construction (discussed in § 2.2). Pragmatically, the reason for the co-occurrence of the reciprocal verb with the reciprocal pronoun has to do with an ambiguity many languages with verbal encoding encounter. Naturally, it is hard to demonstrate this ambiguity in a extinct language, but BAR-ASHER (2009a, pp. 240-242) has demonstrated this in other languages with verbal encoding:

When reciprocity is not expressed by a discontinu-ous construction and there is a plural subject, both a collective and a distributive reading are possible. In other words, while the sentences almost always express events with reciprocal relations, it is not necessary that the sentence describes a single event, i.e. that all the members of the subject-set participate as a collective in one single event. It is possible that each member of the subject-set indeed participates in reciprocal events, but that these events may involve other entities, not neces-sarily present in the sentence. Thus each member of the subject is participating in different events, and the sentence should be read . This is similar

64. Bt stands here as a general term of the T-stems

to the ambiguity of (52) which has two readings, while in most social contexts (53) has only one, due to pragmatic considerations.

(52) John and Mary got marrieda. Mary married John (and John married Mary)b. John married someone (who is not Mary) and

Mary married someone (who is not John)(53) My brother and my sister got married last week.

a. #My brother married my sister last week.b. My brother married someone and also my sister

married someone (else) last week.

Hebrew, Modern Greek, and Turkish wherever a discontinuous reciprocal is possible:

(54) Modern HebrewMPL.PST.kiss-RECP

someone” [not each other, and not necessarily the same person]

(55) Modern GreekDEF Giannis and DEF Mary kiss-RECP-3PL.PST

i. “Giannis and Maria kissed each other”ii. “Both Giannis and Maria had a reciprocal kissing

with someone” [not each other, and not necessarily the same person]

(56) John ve Mary op-us-tu-ler TurkishJohn and Mary kiss-RECP-PST-3PL

i. “John and Mary kissed each other”ii. “Both John and Mary had a reciprocal kissing with

someone” [not each other, and not necessarily the same person]

There are various ways to disambiguate the meaning of these sentences. For example, with the addition of a counting with a distance distributor to the plural subject, the only available interpretation is the distributive reading [the (ii) from each of the above sentences]. According sentences in (57) have only one reading (I bring the evidence from Hebrew, for the data in the other languages see BAR-ASHER 2009a):

(57) Modern Hebrew

MPL.PST.kiss-RECP PL every one

someone [not each other, and not necessarily the same person]”.

A strategy to specify collective reading is to add the pronominal encoding:

(58) MPL.PST.kiss-RECP one with DEF-second

39

Based on this evidence, if indeed this was also the case in Akkadian, the appearance of a detransitivized verb with the one-unit pronoun functions similarly to other languages with a verbal encoding, and makes explicit the reciprocal relation between the members of the subject set. Therefore, this is not an indication of the loss of the verbal encoding of the reciprocity at the later periods, as argued by Buccellati.

As noted it is hard to demonstrate contexts of ambi-guity in a extinct language, but possibly the following example represents a good case were the explicit pronoun could disambiguate the meaning:

(59) =mito-RECP 1PL.PST RECP=and“We fought with each other” (AASOR 16 72:10, Nuzi)/

On the basis of the context of this lawsuit concerning

that it was important to make it explicit that it is not the case that they fought together against another party, but rather that they fought each other. Indeed once it

=mi appears without the reciprocal pronoun as this ambiguity was resolved, indicating that

it could carry the same reciprocal meaning without the pronoun.

As for the high frequency of the transitive verbs with reciprocal pronouns in the later periods, although Buccellati did not provide clear evidence for this claim,65 the impression from surveying the data for this paper is that there is some support for it. If one takes, for example, the uses of the verb “to see”, in the old dialects (and for this matter in SB as well) it is used in the N-Stem with the reciprocal meaning of “seeing each other” or “meet”, as, for example, is the case in example (1a), repeated below:

(60) and you 1PL.PST.meet.RECP“We met, you and I” (OIP 27 15:22, OA).

By contrast, in an example from MB we encounter the use of this verb in the G-stem (the transitive stem) with the reciprocal pronoun :

(61) RECP INJ 1PL.PST.see“Let us see each other” ( 10, p. 3:13 and 15).

However, by no means we do not encounter examples of the T-stems with reciprocal meaning in the later periods – as shown, for example, in some of the

65. For some support for Buccellati’s claim, see PAVIE 2006, pp. 7-11.

examples in this paper (inter alia 2d-e).66 In light of

the reality, the content of the description should be rephrased that in the later periods, in the choice between the verbal and the pronominal reciprocal constructions, the latter was favored. It is possible that this develop-ment is in fact related to the historical development discussed above (§ 3.3), the shift from two-unit pronouns to the one-unit pronoun.

As demonstrated earlier (§ 3.2), the two-unit pro-nouns are available only without an explicit subject. Thus, their use is much more restricted. The one-unit pronoun is available also with explicit subject, and hence it is used more often. Accordingly, it is not the case, as argued by Buccellati, that the loss of the reciprocal meaning of the T-stems motivated the use of the one-unit pronoun The opposite is in fact the case: once the shift to a one unit-pronoun happened, the pronominal encoding of reciprocity became more useful (also with explicit subject). Consequently the use of pronominal encoding became more frequent, and the need of verbal encoding much less.

6. CONCLUSIONS

This paper concentrated on the various reciprocal constructions in Akkadian. Starting with the main dis-tinction between verbal and the pronominal construc-tions (§ 2), and considering the various pronominal strategies (§ 3.1) a reciprocal situation may be repre-sented in one of the following syntactic forms:

1/ A transitive verb with reciprocal pronouns, with either a/ a plural subject with a one-unit pronoun and a

plural verb (3d) or, b/ a singular verb with the two-units pronoun (3a),

and one of the elements of the pronouns function as a subject (a few examples with plural verbs [12a-b] were discussed as well).

2/ A detransitivized verb consisting of either (or a semantic symmetric predicates [41]), with eithera/ a plural verb (2c), orb/ a plural verb with reciprocal pronoun after the

associative preposition (41), or c/ only one set in the subject position and the other set

after the associative preposition (4b, the discon-tinuous construction).

66. From our survey it seems that also in the later periods it

40

This is not to say that every reciprocal situation may be encoded with all these strategies, since the verbal construction seems to be restricted semantically to a certain type of events (§ 4, however, the validity of this claim in the context of Akkadian has been examined in this section).

With regards to the pronominal construction it has been noted that the various constructions are subgroups within a broader strategy of expressions of reciprocity with the regular transitive verb (§ 3.4.1).

Besides the various syntactic distinctions between the one-unit and the two-unit pronominal constructions (§ 3.2), it has been noted that each belong to different periods of in the history of Akkadian (§ 3.3). Considering the history of the pronominal expressions, proposals for the origins of each of the various constructions have been proposed (§ 3.4-6).

The historical development within Akkadian, and the different in syntax between the constructions together with the semantic differences between the verbal and the pronominal encodings (§ 4) were used to explain the fact that verbal construction seems to appear more frequently in the earlier periods and less later (§ 5).

REFERENCES

AMBERBER M.2002

(LINCOM, Studies in Afroasiatic Linguistics), München, Lincom Europa.

BAR-ASHER E.2009a -

, PhD diss., Harvard University.

2009b “Dual pronouns in Semitics and an evaluation of the evidence for their existence in Biblical Hebrew”,

46, pp. 32-49.BAR-ASHER SIEGAL E.A.

forthcoming , in J. TROPPER, J. HUEHNERGARD,

L. KOGAN, D. NICOLAE, J.-P. VITA (dir.),

, Münster, Ugarit-Verlag (JBA).forthcoming “Diachronic syntactic studies in the Hebrew

pronominal reciprocal constructions”, in C. MILLER, Z. ZEVIT (dir.), , Winona Lake, Eisenbrauns (Hebrew).

BEHRENS L.2007 “Backgrounding and suppression of reciprocal

Participants: A cross-linguistic study”,31, pp. 327-408.

BORGER R.1996

, Wiesbaden, Harrassowitz.BUCCELLATI G.

1996 , Wiesbaden, Harrassowitz.

BUTTS A.M.2010 “The etymology and derivation of the Syriac

adverbial ending *- ”, 69, pp. 79-86.

CORBETT G.G.2006 Agreement, Cambridge, Cambridge University Press.

DALRYMPLE M., MAKOTO K., YOOKYUNG K., MCHOMBO S., STANLEY P.

1998 “Reciprocal expressions and the concept of recipro-city”, 21, pp. 159-220.

DELITZSCH Fr.1889

, Berlin, H. Reuther.DIMITRIADIS A.

2004 “Discontinuous reciprocals”, Unpublished ms., UiL-OTS, December.

EVANS N., GABY A., NORDLINGER R.2007 “Valency mismatches and the coding of reciprocity

in Australian Languages”, 11, pp. 541-97.

2007a “Reciprocals grammar-writing and typology”, Revue de linguistique roumaine, II (4), pp.479-506.

FRAJZYNGIER Z., CURL T.S. (dir.),1999 , Amsterdam,

John Benjamins.GELB I.J.

1957 “Notes on von Soden’s Grammar of Akkadian”, 12, pp. 93-111.

GELB I.J. LANDSBERGER B., et alii1956 Chicago, The Oriental

Institute.GLINERT L.

1989 , Cambridge, Cambridge University Press.

GOLDENBERG G.1991 “‘Oneself’, ‘one’s own’ and ‘one another’ in

Amharic”, in A.S. KAYE (dir.), , Wiesbaden, Otto Harras-

sowitz, pp. 531-549.GRAYSON A.K.

1972 , 1, Wiesbaden, Harras-sowitz.

GRONEBERG B.1989 “Reduplications of Consonants and R-Stems”,

83, pp. 27-34.HASPELMATH M.

2007 “Further remarks on reciprocal constructions”, in NEDJALKOV 2007, pp. 2087-2115.

41

HASSELBACH R.2007 “External plural markers in Semitic: A new assess-

ment”, in C.L. MILLER (dir.),

(Studies in Ancient Oriental Civilization, 60), Chicago, The Oriental Institute, pp. 123-138.

HECKER K.1968 (Analecta Orienta-

HOPPER P.J., THOMPSON S.A.1980 “Transitivity in grammar and discourse”, 56, pp. 251-299.

JAY J.2009 “Reciprocal constructions in Biblical Hebrew”,

, 3/1.KEMMER S.

1993 , Amsterdam, John Benjamins.

KHAN G.2004

, Leiden, Brill.KÖNIG E., GAST V. (dir.)

2008 -, Berlin, Mouton de Gruyter.

KÖNIG E., KOKUTANI Sh.2006 “Towards a typology of reciprocal constructions:

Focus on German and Japanese”, 44, pp. 271-302.

KOUWENBERG N.J.C.1997 (Studia Semitica

Neerlandica, 32), Assen (Netherlands), Van Gorcum.KREMERS J.

1997 nafs and

, M.A. thesis, University of Nijmegen.LIPI SKI E.

1997 (Orientalia Lovaniensia Analecta, 80),

Leuven, Peeters. LOESOV S.

2004 “T-Perfect in Old Babylonian: Current debate and a thesis”, 1, pp. 83-181.

MALONEY J.F.1982

, Unpublished Dissertation, Harvard University.

MASLOVA E.2000 “Reciprocals and set construal”, in FRAJZYNGIER,

CURL 2000, pp.161-178.MEYER L. DE

1962 , Leiden, Brill.

MUTZAFI H.2004

(Semitica Viva, 32), Wiesbaden, Harrassowitz.

NEDJALKOV V.P. (dir.)2007 , 5 vols, Amsterdam,

John Benjamins.

framework, and related issues”, in NEDJALKOV 2007, pp. 3-114.

2007b “Encoding of the reciprocal meaning”, in NEDJAL-KOV 2007, pp. 147-208.

2007c “Polysemy of reciprocal markers”, in NEDJALKOV 2007, pp. 231-334.

PAVIE N.2006 “Questions relatives au système des

de l’Akkadien”, M.A. Thesis, Hebrew University of Jerusalem.

RUBIN A.D.2005 (Harvard

Semitic Series, 57), Winona Lake, Eisenbrauns.2010 , Boston, Brill.

SALONEN E.1962 -

(Studia Orientalia, 27/1), Helsinki.SAPIR E.

1931

(65), Boston, Ameri-

can Academy of Arts and Sciences.SILONI T.

2001 “Reciprocal verbs”, 17 (available

at http://atar.mscc.huji.ac.il/_english/IATL/17)2002 “Active Lexicon”, 28,

pp. 383-400.2008 “The syntax of reciprocal verbs: An overview”, in

E. KÖNIG, V. GAST (dir.), ,

Berlin, Mouton de Gruyter, pp. 451-498.SODEN W. VON

1933 “Der hymnisch-epische Dialekt des Akkadischen”,

[ZA] 41, pp. 90-183.1952 , Rome,

1965-1981 , Wiesbaden, Harrassowitz.

STRECK M.P.2003 (Alter

Orient und Alter Testament, 303), Münster, Ugarit-Verlag.

42

STROOMER H.1999

, Wiesbaden, Harrassowitz.TESTEN D.

1998 “The derivational role of the Semitic N-Stem”, ZA 88, pp. 127-145.

VOIGT R.2001 “Semitische Verwandtschaftstermini”, in A. ZABOR-

SKI (dir.), , Wiesbaden,

Harrassowitz, pp. 205-218.WINTER Y.

2007 “Reciprocals and the lexical semantics of predicates”, Paper presented at

, FU Berlin, November 30-December 2, 2007.